Fig. 1

The workflow for data matrix construction. a–e: the workflow for obtaining data matrix. a: the coding sequence (CDS) of A. thaliana was used as template for mapping. RNA-Seq reads were translated into amino acid sequences and mapped to the template using BLAT in protein space; b: Continuously mapped reads were retained after passing minimal BLAT mapping score (see Methods), and exact read-mapped regions on the template were then extracted. c: UCS, CS and AS were determined by calculating the nucleotide frequency at each site based on the mapping result (see Methods); d: The codons were extracted from CS using sliding windows. e: linking retained codons for each CDS, CS data matrix was then built by concatenating retained codons from all CDS for ML method. (Abbreviations: UCS: uncovered site, CS: consensus site, AS: ambiguous site.)